Markit is pleased to submit the following comments to ESMA in response to its Discussion Paper on Benchmarks Regulation (the DP ).

ESMA 103 rue de Grenelle Paris 75345 Submitted online via www.esma.europa.eu London, March 31 st 2016 ESMA Discussion Paper on Benchmarks Regulation Dear Sirs, Markit is pleased to submit the following comments to ESMA in response to its Discussion Paper on Benchmarks Regulation (the DP ). Markit 1 is a leading global diversified provider of financial information services. 2 Founded in 2003, we employ over 4,000 people in 11 countries and our shares are listed on Nasdaq (ticker: MRKT). Markit has been actively and constructively engaged in the debate about regulatory reform in financial markets, including topics such as the implementation of the G20 commitments for OTC derivatives and the design of a regulatory regime for benchmarks. Over the past years, we have submitted more than 140 comment letters to regulatory authorities around the world and have participated in numerous roundtables. Introduction Indices and benchmarks play an important role in enhancing transparency, liquidity and access in financial markets around the globe, they contribute to broadening the base of investable assets in the EU and are a crucial element for the success of the CMU project. We welcome the EU Benchmark Regulation 3 ( the Regulation ) as an important framework to protect investors and restore confidence in benchmarks. In the context of this DP we recommend ESMA ensure that the final rules are in line with the level 1 text and they are proportionate as well as workable so they do not chill innovation in this important sector. Specifically, our main recommendations are: In its determinations in relation to the scope of the Regulation, ESMA should be guided by the significance of conflicts of interest and the potential for investor detriment. Specifically, the term published or made available to the public as part of the definition of benchmark should be understood to mean any distribution of the benchmark to one or more external parties. Anything narrower would fail to address potential conflicts of interest and open opportunities for regulatory arbitrage. Similarly, where a firm combines benchmarks, it should be deemed an administrator where discretion is involved and/or significant conflicts exist. 1 See www.markit.com for more details. 2 We provide products and services that enhance transparency, reduce risk and improve operational efficiency of financial market activities. Our customers include banks, hedge funds, asset managers, central banks, regulators, auditors, fund administrators and insurance companies. By setting common standards and facilitating market participants compliance with various regulatory requirements, many of our services help level the playing field between small and large firms and foster a competitive marketplace. 3 Like ESMA in the DP, we refer in this letter to the Council text agreed in December 2015 http://data.consilium.europa.eu/doc/document/st-14985-2015-init/en/pdf

The inherent conflicts in the provision and use of benchmarks can be effectively mitigated through independent oversight or administration. In this case, the party which controls the methodology for a benchmark and the administrator will be separate from the benchmark owner or user. We believe that this approach has been contemplated in the IOSCO Principles for Financial Benchmarks and recommend ESMA explicitly recognise it as a best practice approach to mitigating conflicts of interest and as an alternative to divestiture of benchmarks by providers who lack the necessary controls. We agree that verification of submissions to benchmarks should take place where possible and appropriate. However, this will be best achieved by ensuring benchmark administrators can request the relevant information used to generate a submission from contributors when they see a need, rather than requiring them to obtain all of the relevant data supporting all submissions on an ongoing basis which would create significant logistical and commercial challenges. ESMA should recognise that measuring the notional amount referenced to benchmarks will be challenging and requirements on benchmark administrators will need to reflect their ability to collect the data. When benchmarks are used to determine the performance of an investment fund, they should be fully in scope of the regulation to ensure adequate protection of investors. However, to the extent a benchmark is used to provide only a relative measure of performance and has in itself no impact on the value of the fund the relevant notional should not count towards whether it is considered critical or significant. Responses to ESMA s questions 2) Definitions (A3) Q1: Do you agree that an index s characteristic of being made available to the public should be defined in an open manner, possibly reflecting the current channels and modalities of publication of existing benchmarks, in order not to unduly restrict the number of benchmarks in scope? The Regulation applies to Benchmarks that rely on an index that is published or made available to the public. 4 We agree with ESMA that made available to the public ought to be defined in an open manner, also because the level 1 text made it very clear that the Regulation should be as broad as necessary to mitigate conflicts of interest and potential investor detriment. 5 We understand that the original intention for the Regulation to apply only to indices that were made available to the public was to ensure that indices that were produced solely for the use of in house research would not be captured, rather than generally limit the scope of the Regulation to only indices with broad publication. In this determination ESMA should therefore be guided by when potential conflicts of interest exist in the administration of the benchmark and ensure that all such cases are brought into the scope of the Regulation. It will be important that administrators cannot place themselves outside the Regulation by simply restricting the distribution of their indices in some manner (for example by limiting the number of users, by imposing a fee for access to the benchmark, or by disseminating indices via a password protected website). While any setting of a specific number of recipients would open up opportunities for such regulatory arbitrage, ESMA should also consider that, even where an index is distributed to a single asset manager, the risk of it not being administered appropriately could impact a large number of its investors. The final rules should therefore not create any artificial boundaries (for example, by specifying a certain minimum number of recipients or types of subscriptions) in relation to the definition of made available to the 4 Article 3(1)(1)(a) 5 Recital 8 states that The scope of this Regulation should be as broad as necessary to create a preventive regulatory framework. 2

public. We believe the simplest and most effective approach to this issue would be to define made available to the public as situations where the administrator makes the index available to one or more external parties. However, to not stifle innovation in the indices and benchmark sector, we recommend ESMA exempt benchmarks from this definition that are disseminated only to a limited number of users on a trial basis or while being under development. Q2: Do you have any proposals on which aspects of the publication process of an index should be considered in order for it to be deemed as having made the index available to the public, for the purpose of the BMR? As stated above, we believe the level 1 is clear that made available to the public ought to be defined in an open manner and the scope of the Regulation should be as broad as necessary to mitigate conflicts of interest and prevent potential investor detriment. The best way to ensure this would be for an index to be deemed as having been made available to public if it is provided to one or more external parties. The Regulation aimed to introduce a common framework to ensure accuracy and integrity of indices used as benchmarks in the interest of consumer and investor protection. To achieve these objectives it should consider that there are risks of detriment applicable to each and every benchmark where ultimately investors can be affected by low standards or conflicts of interest in the administration of the benchmark. ESMA s proposal to map the currently used publication channels and costs and ease of access and use those as basis for the final rules 6 therefore seems inconsistent with the need to ensure a broad scope for the Regulation; it might result in creating arbitrary boundaries for the scope of the Regulation and open up the potential for regulatory arbitrage. For example, if by requiring the registration of a user on a webpage to access an index or by distributing it to only a limited number of users an index provider could move a benchmark out of scope of the Regulation, it could easily introduce such restrictions and the Regulation would not succeed in reducing risks or conflict of interest for the investors exposed to this benchmark. Combinations of benchmarks Similarly, ESMA should also carefully consider its approach to determining when combinations of benchmarks should be classified as use of a benchmark. As set out in Recital 3a, where a user creates a composite benchmark by blending or combining multiple benchmarks in a manner that does not involve discretion, it is using the underlying benchmarks. ESMA should clearly distinguish this from situations where the user does exercise discretion in combining benchmarks and/or has an interest in the level of the resulting benchmark. Where a firm modifies or otherwise changes one or several existing benchmark by using discretion to form a derived benchmark, it should be considered to be the administrator of the derived benchmark. For example, changes to the proportion of benchmarks used in the creation of a combination of benchmarks may involve discretion. We recommend ESMA ensure the final rules are clear that where discretion is used in combining benchmarks it is regarded as administration. Uncertainty in this area could lead to difficulties for firms interpreting the Regulation. In this context, ESMA should note that the FCA recently reported on the findings of its thematic review on benchmarking activities. Specifically, it found that at least one bank had not given full consideration to conflicts of interest that could arise when traders provided a rebalanced basket of indices or updated prices for a credit/equity strategy to clients. 7 A lack of clarity about the impact of the use of discretion in combining benchmarks could lead to asset managers experiencing similar problems. 6 Pg. 11, Par. 12 and 15 3

Q3: Do you agree with ESMA s proposal to align the administering arrangements for determining a benchmark with the IOSCO principle on the overall responsibility of the administrator? Which other characteristics/activities would you regard as covered by Article 3(1) point 3(a)? We agree with ESMA that the administering arrangements should be aligned with the IOSCO Principles for Financial Benchmarks. However, we are concerned that the DP sets out that administration includes development, determination and dissemination, operation and governance 8 of a benchmark. We urge ESMA to ensure that any definition of the responsibilities of an administrator, particularly in relation to the development of the benchmark, does not preclude the owner of the intellectual property of the benchmark from effectively engaging an independent administrator. This is because independent benchmark administrators can help parties that develop indices to mitigate conflicts of interest they might be exposed to and benefit from the expertise that such specialist services can bring. We therefore recommend ESMA ensures that its final rules do not unintentionally discourage such approach by suggesting that the benchmark administrator needs to have been involved in the original development of the benchmark. This could be achieved by clarifying that an entity could be classified as administrator regardless of whether or not it owns the intellectual property of the benchmark or has been involved in its development. This clarification would not in any case diminish the responsibility of the administrator going forward. Q4: Do you agree with ESMA s proposal for a definition of issuance of a financial instrument? Are there additional aspects that this definition should cover? We agree that the concept of issuance of financial instruments should not be limited to securities and should extend to financial instruments that are created for trading also in execution or trading venues other than regulated markets. 3) Oversight function requirements (A5a) Q6: Do you agree with the proposed list of appropriate governance arrangements for the oversight function? Would you propose any additional structure or changes to the proposed structures? We are generally supportive of the principles underlying ESMA s proposed list. In particular we support the possibility for a natural person to perform an oversight function and the possibility for an oversight function to cover a number of benchmarks as these approaches will allow for a proportionate and workable implementation. 9 However, we recommend ESMA apply a greater focus on areas where conflicts of interest could appear in governance arrangements and how they can be mitigated. Specifically, ESMA should neither require nor forbid certain arrangements (for example, including representatives of contributors or of providers of data sources used to determine the benchmark in the oversight function) as any overly prescriptive approach could lead to increased conflicts of interest for some benchmarks. ESMA outlines that, where conflicts of interest arise on the level of the administrator, independent governance structures may be required by the authorities. We believe this is generally a valid approach but recommend ESMA explicitly reflects in the final standards the ability for the owners of benchmarks to use independent administrators. For example, in the UK the FCA has actively examined management of the risks around benchmarks and conflicts of interest. 10 Also, several major owners of benchmarks have engaged independent administrators for the day-to-day administration of their indices, allowing users of their benchmarks to benefit 8 Pg. 14, Par. 25 9 For example in Pg. 19-21, Par. 43, 46 10 For example through its Thematic Review of Financial Benchmarks TR15/11 - www.fca.org.uk/static/documents/thematicreviews/tr15-11.pdf 4

from their specific and independent index administration expertise, mitigate conflicts of interest, and to satisfy regulatory best practice. 11 Independent benchmark administration We note that the arrangements suggested in the DP do not explicitly recognise the use of third party index administrators which could lead to confusion about whether their use is accepted to manage the conflicts of interest identified. Independent benchmark administrators have no commercial interest in the benchmark (other than its quality and its representativeness) and this should be reflected in the requirements for independent benchmarks: for example, conflicts of interest can lie with the ability to change the methodology, but when an owner of the benchmark transfers control of the methodology to an independent third party, the conflicts of interest are mitigated while maintaining ownership of the benchmark. In several sections of the DP ESMA seems to refer to independent benchmark administrators by referring to entities that are dedicated to providing benchmarks alone. 12 We believe that the use of such definition would be misleading and might unintentionally favour firms that happen to provide only provide indices rather than many of the more relevant independent benchmark administrators. This is because in reality many independent benchmark administrators offer a range of services, including some that might not be related to indices and benchmarks. We therefore urge ESMA to recognize that the most important factor determining the independence of administrators is whether or not they take any positions in financial instruments that reference or are themselves users of the benchmarks they administer. We hence recommend ESMA focus on potential conflicts of interest and whether the benchmark administrator could have a financial interest in the outcome of the Benchmark and ensure that its final standards do not disadvantage independent administrators just because they perform some further, but unrelated activities. On this basis we recommend that ESMA: Focus on the existence of conflicts of interest, especially whether the administrator or committee members have a commercial interest in the level of the benchmarks administered or not; Recognise that the use of independent administrators is a valid method for the owners of the benchmark who might take positions in financial instruments to mitigate the risks around conflicts of interest; and Recognise that an administrator can be independent even if it performs other activities beyond benchmark administration, as long as it does not take positions in financial instruments that reference the benchmarks it administers. Q7: Do you believe these proposals sufficiently address the needs of all types of benchmarks and administrators? If not, what characteristics do such benchmarks have that would need to be addressed in the proposals? As stated in our answer to question 6, we encourage ESMA to specifically recognise independent administration as a valid method of mitigating the most severe conflicts of interest. Any governance requirements set in level 2 for administrators should be proportionate to the potential risks inherent in those arrangements. ESMA should also note that there may be cases where contributors to a benchmark are also users of that benchmark (for example proprietary benchmarks administered by banks or performance benchmarks administered by asset managers). Because of these situations, ESMA should neither require nor forbid certain arrangements and provide the administrator with the final responsibility in determining which composition of the oversight function is best suited to mitigate the conflicts of interest specific to its benchmark. 11 See www.markit.com/company/media-centre/ubs-selects-markit-for-index-administration 12 Pg. 73, Par. 228 and Pg. 76 Par. 247 5

Q8: To the extent that you provide benchmarks, do you have in place a pre-existing committee, introduced through other EU legislation, or otherwise, which could satisfy the requirements of an oversight function under Article 5a? Please describe the structure of the committee and the reasons for establishing it. Markit develops and administers a range of proprietary indices covering loans, bonds, credit default swaps, structured finance and economic indicators. We are also an independent administrator for a number of customer-owned benchmarks. In this role, we control the customer benchmark methodology and oversee all submission, determination and distribution processes. We administer both our own and our customers benchmarks in accordance with the IOSCO Principles for Financial Benchmarks. Our compliance framework includes a governance and oversight function designed to reconcile regulatory compliance with daily business management and industry expertise. Specifically, our oversight and governance structure includes the following key components: Index Management team: a supervisory group of internal business experts that governs all of Markit s proprietary indices and is accountable for the overall operation of Markit s benchmarks. Index Administration Committee: a core group of individuals who hold a range of roles in the customerowned benchmark administration business and provides oversight and supervision of these benchmarks. Index Advisory Committees: committees comprised of stakeholders and other industry participants that provide expert external feedback for matters related to the rules and administration of specific indices. The objective of these consultations is to solicit feedback which can assist the administrator in the creation of the highest quality standards for its indices. Benchmark Oversight Committee: a committee composed of participants outside of Markit s benchmark administration businesses including members of Markit s legal, compliance, finance, risk, information security and executive management teams. The Benchmark Oversight Committee is responsible for reviewing and scrutinising all aspects of Markit s compliance with its Administrator Code of Conduct and the IOSCO Principles for Financial Benchmarks. Q9: Do you agree that an administrator could establish one oversight function for all the benchmarks it provides? Do you think it is appropriate for an administrator to have multiple oversight functions where it provides benchmarks that have different methodologies, users or seek to measure very different markets or economic realities? We agree with ESMA that there might be situations where a single benchmark oversight committee could effectively oversee all of the benchmarks of the administrator. If there was a requirement to establish separate oversight committees for each benchmark, it might lead to significant duplication and unnecessary cost as the tasks and likely composition of the committee might be very similar, if not identical, for many benchmarks or families of benchmarks. We therefore strongly believe that the administrator should be permitted to make this decision on a case by case basis depending on the nature of the benchmarks it administers. Markit s Benchmark Oversight Committee has overall responsibility for the effective scrutiny of Markit s benchmark administration processes. The various Index Management Boards and Index Administration Committees provide information and report to the Benchmark Oversight Committee on a regular basis in connection with its responsibility to review and challenge all aspects of the benchmark determination process. Q10: If an administrator provides more than one critical benchmark, do you support the approach of one oversight function exercising oversight over all the critical benchmarks? Do you think it is necessary for an oversight function to have sub-functions, to account for the different needs of different types of benchmarks? As stated above, we believe the oversight committee s role includes ensuring effective governance and control for all the benchmarks or families of benchmarks under its authority. Part of this task will be to ensure that appropriate functions specific to each benchmark are operating effectively. We therefore believe the administrator 6

should be permitted to implement a single oversight function where it can demonstrate that such approach is appropriate given the nature, number and complexity of the benchmarks it administers. Q11: Where an administrator provides critical benchmarks and significant or non-significant benchmarks, do you think it should establish different oversight functions depending on the nature, scale and complexity of the critical benchmarks versus the significant or non-significant benchmarks? As stated above, we believe the oversight committee s role includes ensuring effective governance and control for all the benchmarks or families of benchmarks under its authority. Part of this would be to ensure that appropriate functions specific to each benchmark are operating effectively. Therefore we would support the use of a single oversight function where appropriate. Furthermore, the use of a single oversight committee covering all categories of benchmarks would ensure that all benchmarks benefited from the same high level of expertise, experience and consistent approach. Q12: In which cases would you agree that contributors should be prevented from participating in oversight committees? When establishing the oversight committee the administrator should ensure that its members are sufficiently independent whilst also having the necessary expertise to effectively challenge the management of the administrator. We believe it should be left to the administrator to decide whether or not to admit contributors to the oversight committee as long as it fully considers and mitigates potential conflicts of interest in relation to this decision. In this context, ESMA should note that contributors to benchmarks may be subject to significant conflicts of interest, especially when they are also users of the benchmark. However, contributors actively participate in markets that are represented by the benchmarks and they hence often have the necessary expertise to play a useful role as members of the oversight committee. ESMA also recognises this when it states that where a benchmark is based on submissions and the oversight function takes the form of a committee, it may be important for representatives of the contributors to participate in the [oversight] committee. 13 We therefore recommend that ESMA neither require nor forbid certain arrangements in this respect but provide the administrator with the final responsibility in determining which composition of the oversight function is best suited to providing the necessary expertise while mitigating conflicts of interest. However, one should not admit to the oversight committee those specific individuals that themselves make submissions to the administrator or enter into transactions in benchmark related instruments as this would pose unacceptable conflicts of interest. In contrast, an individual that is employed by the contributor to oversee the submissions will be well positioned to provide valuable expertise whilst not being exposed to conflicts of interest that are not manageable. Q13: Do you foresee additional costs to your business or, if you are not an administrator, to the business of others resulting from the establishment of multiple oversight functions in connection with the different businesses performed and/or the different nature, scale and type of benchmarks provided? Please describe the nature, and where possible provide estimates, of these costs. As stated above, we support the ability for the administrator to establish a single oversight function if it believes it would be best suited given the nature of the benchmarks it administers. Q14: Do you agree that, in all cases, an oversight function should not be responsible for overseeing the business decisions of the management body? 13 Pg. 21, Par. 45 7

We agree that, in all cases, an oversight function should not be responsible for overseeing the business decisions of the management body. This is because the key function of the oversight committee is to provide effective challenge to the day-to-day administration of the benchmark. We believe that overseeing business decisions would expose the oversight committee to an unmanageable conflict of interest and question the independence of the function. For example, the oversight function could conclude that the administrator needs additional sources of data to ensure that a particular benchmark is representative of the market it covers but should not be involved in negotiating contracts with data providers or contributors in securing the data in question. Q15: Do you support the proposed positioning of the oversight function of an administrator? If not, please explain your reasons why this positioning may not be appropriate. We agree with ESMA that an oversight function should be embedded within an administrator s organisation to operate effectively and that internal members of the administrator s organisation should be represented in the oversight committee because they have the same knowledge and access to the confidential business details of a benchmark administrator in order to influence and effectively challenge the management body. We further agree that any conflict of interest for internal members of the oversight committee can be mitigated by ensuring operational separation from the parts of the administrator structure dealing with the provision of benchmarks. 14 Q16: Do you have any additional comments with regard to the procedures for the oversight function as well as the composition and positioning of the oversight function within an administrator s organisation? ESMA suggested that the oversight committee should operate as a consultative body of the administrator, interacting directly with the management body while it, at the same time, independently reports any misconduct by contributors or administrators, of which it becomes aware, to the relevant competent authorities. 15 We generally agree with ESMA s assessment of the roles and responsibilities of the oversight committee. For the oversight committee to be able to discharge its duties effectively it may be beneficial to consider active participation from all stakeholders including the contributors, the administrator and the users of the benchmark. In doing so, the composition of the oversight function should be adequately distributed across different functions within these organisations to ensure that the individual members of the committee are able to impose sufficient scrutiny on the activities of the administrator and its adherence to the Regulation. In addition, administrators should choose the persons from these organisations to minimize conflicts of interest. For example, the oversight committee should generally not include any persons making submissions to or taking positions in the benchmark. Q17: Do you agree with the proposed list of elements of procedures required for all oversight functions? Should different procedures be employed for different types of benchmarks? We generally support the list of elements of procedures that ESMA proposed to be required for all oversight functions. However, we would be concerned if additional requirements were prescribed or forbidden by ESMA. For instance, we would question the value of requiring the administrator to publish the names of individual members of any board or committee charged with the oversight function and trust a description of the functions represented in the oversight functions provides a sufficient detailed summary of membership. As stated above, 14 Pg. 22, Par. 47 15 Pg. 22, Par. 49 8

we believe it should be the responsibility of the oversight committee and the administrator to ensure that appropriate procedures are in place. Q18: Do you agree with the proposed treatment of conflicts of interest arising from the composition of an oversight function? Have you identified any additional conflicts which ESMA should consider in drafting the RTS? Please refer to our response to Question 16. 4) Input Data (A7) Q19: Do you agree with the list of records to be kept by the administrator for input data verification? If not, please specify which information is superfluous / which additional information is needed and why. We believe that ESMA s proposed approach to the record keeping requirements for administrators could be highly problematic, disproportionate and duplicative, and would cause significant commercial and logistical challenges. We recommend that administrators establish the right to request relevant data underlying submissions from submitters to be in a position to verify and/or challenge how any submission was determined. We agree with ESMA that the costs of the actual storage of the records would not be the main issue in this context. However, key areas of concern are the difficulty of arranging for the transmission of such data and the confidentiality of such data (which often represent trade secrets and competitive intelligence) resulting in further exacerbating the disincentives for submissions. Specifically, any regular submission of all of the relevant data would be highly problematic for the following reasons: ESMA suggests the administrator would need to obtain the names and role(s) of the individuals responsible for submissions and approval. 16 However, ESMA should note that contributors will be reluctant to provide this personal data while it is unclear how this would support the stated goal of being able to reconstruct a given contribution based on recorded information. Furthermore, holding details of individuals would require administrators to record non-public personal data that could prove problematic in the context of data protection rules, increase the burden and provide further disincentive for contributors to continue their contributions. 17 Requiring administrators to keep records of material transactions or market data which were deliberately excluded from contribution would also be extremely problematic. This is because administrators do not have access to the transactions or positions of their contributors and contributors would be highly unlikely to provide such confidential information to them on a systematic basis. Requiring administrators to keep records of deviations from procedures and practices governing contributions as well as, in case of non-transaction data, data used, underlying variables, criteria, and assumptions may require a substantial IT system and control change for the contributor to be able to deliver this piece of information to the administrator in an efficient and systematic way. We welcome ESMA s proposal for only relevant communications between the parties involved (i.e., contributors, approvers and calculation agents) to be recorded by the administrator. 18 We believe such an approach could make this requirement more manageable, especially for smaller administrators. In its final proposals ESMA should further allow administrators to assess the relevance of communication to be stored. ESMA should also note that the requirement to record substantial exposures of individual traders or trading 16 Pg. 27, Par.70 17 Pg. 26,27, Par. 67 18 Pg. 27, Par.70 9

desks to benchmark related instruments, as well as changes therein 19 is duplicative to the requirement of Article 7(3a) which requires administrators to exercise additional scrutiny on submissions made from a front office function. ESMA should also consider whether the level of information required to be stored could be made more proportionate to the nature and number of the contributors. For example, for a benchmark that only covers a single variable and is based on contributions from only a few submitters it may be more appropriate to keep detailed records. In contrast, where the determination of the benchmark relies on contributions from a large number of submitters for numerous instruments, not only would the burden of record keeping be far greater but also keeping detailed records of each and every contribution seems unnecessary as outliers and anomalies could be more easily identified based on the information received. Finally, we encourage ESMA to clarify exactly which data the requirements would apply to. Specifically, we believe it would be disproportionate to require benchmark administrators to keep detailed records of data that is readily available or originally produced for purposes other than the benchmark, such as valuation for instance. Q20: Do you agree that, for the information to be transmitted to the administrator in view of ensuring the verifiability of input data, weekly transmission is sufficient? Would you instead consider it appropriate to leave the frequency of transmission to be defined by the administrator (i.e. in the code of conduct)? ESMA identified three key processes to be used by administrators to scrutinize their input data, namely evaluation, validation and verification. 20 While evaluation and validation are dependent on the input data submitted by the contributor, verification would consist of an in-depth examination of questionable input data and the process of its creation. 21 The administrator would be responsible for identifying questionable data and would do so based on the benchmark methodology (which would also guide the frequency at which this was done). We urge ESMA to not require administrators to request the data needed for verification at any regular interval. Instead, the administrator should be provided with the discretion to decide on the data it needed for verification and whether and when to require its provision by contributors. We believe that imposing a certain frequency of the provision of data needed for verification, for example weekly, would not be appropriate or allow to reflect differences in benchmark inputs and construction. Furthermore, ESMA should consider that establishing overly onerous verification requirements which required contributors to submit all underlying data on a regular basis would likely result in them ceasing submissions to benchmarks, which might often cause their cessation. We are confident that the regulatory objective of verification of input data can be achieved by ensuring that administrators can request all relevant data they need for verification from their submitters. Administrators rights to request the data should be clearly defined in the Submitter Code of Conduct which would require contributors to retain the relevant records required to verify input data and make it available to the administrator on request. The final standards should allow the administrator to design the provisions of the code of conduct in this respect to reflect the nature of the benchmark and underlying submissions. 22 Q21: Do you agree with the concept of appropriateness as elaborated in this section? 19 Pg. 28, Par.70 20 Pg. 31, Schematic representation of input data scrutiny 21 Pg. 30, Par. 80 22 We disagree with ESMA s view that verifiability of input data is not amenable to being implemented differentially Pg. 35, Par. 103. Please see our response to Q. 30 10

We generally agree with the concept of appropriateness of input data if it is compatible with the benchmark methodology. We agree with ESMA s assertion that on a daily basis, appropriateness of input data cannot be verified except by comparison with the criteria put forward in the methodology. 23 However, in cases where non-transaction data was provided, ESMA asks whether the use of such data [is] justified in light of the methodology and proposes that in such cases, ensuring appropriateness requires that the administrator knows the underlying sources and assumptions on which a contributor bases an estimate. 24 We believe that both of these proposals are problematic and recommend ESMA reconsider them. Specifically, ESMA should note that, where non-transaction data is received as a submission, the underlying sources and assumptions on which a contributor bases its submission often consist of proprietary and sensitive information. In many cases, contributors would therefore rather cease submissions to non-critical benchmarks than sharing this information with administrators. Similarly, for cases where expert judgement was relied upon and the use of such judgement [was] documented it would be difficult for an administrator to assess whether the expert judgement was based on reasonable criteria. 25 We therefore propose that, in cases where no transaction data is available or expert judgment was used by submitters, instead of requiring the administrator to seek justification from contributors on the basis of individual submissions, it should monitor contributors compliance with the general submission framework as defined in its Submitter Code of Conduct. In addition, the administrator should require further information from submitters in cases where input data appears questionable or seems to be in breach of the benchmark methodology as defined by the Code of Conduct. Q22: Do you see any other checks an administrator could use to verify the appropriateness of input data? We routinely apply a multitude of checks for the input data underlying our various indices as well as our pricing and valuation services. Such checks are designed to verify the appropriateness of the input data received on the basis of quantitative analysis, including checks for staleness, consistency for individual contributors as well as between contributors, and comparisons between different data sources for the same or similar financial instruments. We monitor the effectiveness of the verification techniques we employ on an ongoing basis and adjust them where justified by changes in underlying market conditions, liquidity, and/or market structure. Q23: Would you consider it useful that the administrator maintains records of the analyses performed to evaluate the appropriateness of input data? Our experience as index administrator has shown that the creation of a robust index methodology that is complemented by appropriate and representative input data is the cornerstone of successful index determinations. The appropriateness of input data is determined by comparison with the criteria put forward in the methodology. 26 Therefore, to demonstrate that appropriate input data has been used in the benchmark determination process the administrator should maintain records of the analysis it has performed to evaluate the appropriateness of input data. Q24: Do you see other possible measures to ensure verifiability of input data? ESMA should be aware of the challenges to verifying input data depending on the complexity of the benchmark determination process as well as the number of underlying instruments and contributors. 23 Pg 28, Par. 75 24 Pg 28, Par. 75 25 Pg 29, Par. 75 26 Pg.28, Par. 75 11

We believe that, for a critical benchmark such as LIBOR, it would be reasonable to require the administrator to collect and store complete records to verify the submissions, given both the systemic importance of the benchmark and the limited number of contributors making submissions. In contrast, establishing a similar requirement for the administrator of a non-critical benchmark with thousands of underlyings 27 and numerous contributions for each one of these underlyings is very onerous both for the administrator and contributors without delivering commensurate benefits. As a practical approach we therefore recommend that contributors should be required to store all relevant records 28 while the administrator could request these records in cases where the input data provided does not satisfy its evaluation and validation checks. 29 Q25: Do you agree with the identification of the concepts and underpinning activities of evaluation, validation and verifiability, as used in this section? We believe that the concepts of evaluation, validation and verification as explained by ESMA in the DP are generally well established. Markit employs these techniques in the administration of many of its indices to ensure the robustness of its benchmark determination process and in compliance with the IOSCO Principles for Financial Benchmarks. However, we are concerned about ESMA s suggestion that administrators should verify input data where submissions are in breach of the evaluation process and this should be done before the benchmark level is determined. ESMA should note that, for numerous benchmarks, submissions of input data are made in an automated manner which makes any verification of input data pre-determination extremely difficult to achieve. We therefore recommend that administrators, while retaining the ability to intervene pre-determination where necessary, 30 are permitted to apply automated checks after data submission to be able to identify any inaccurate input data. In these situations, the administrator could perform verification of input data mostly in the validation phase (i.e. post-determination) as long as this is sufficient to ensure the robustness of the benchmark. Q30: Do you agree that where expert judgement is relied on and/or discretion is used additional appropriate measures to ensure verifiability of input data should be imposed? If not, please specify examples and reasons why you disagree. We believe it is important for administrators and contributors to benchmarks alike to distinguish between nontransaction data and input data that rely on expert judgement since the proposals emphasise the need for additional verifiability requirements for such data. We believe that the decision about the use of additional appropriate measures for verification should be left with the administrator as it is best positioned to determine what is needed given the nature of the input data and degree of expert judgment used. On a general note, we recommend ESMA clarify what is meant by input data [that] relies on expert judgement. 31 We suggest that such a contribution be understood as an activity conducted by or on behalf of the submitter and for the purpose of creating data solely for submission into the benchmark determination process. We further suggest that data which firms are required to create on an on-going basis, such as indicative or committed quotes, be classified as non-transactional data and not expert judgement. This is because such data is usually created for a different purpose by persons from the contributing firm that are not the submitters. 32 Q31: Do you agree to the list of criteria that can justify differentiation? If not, please specify why you disagree. 27 For example, as of 24 th March 2016 the Markit iboxx Overall Index has more than 3,200 constituents. 28 As mentioned on pages 27/28, Par. 70 (Input data) 29 As enumerated by ESMA in Par.78, 79 (Pg 30, Input data) 30 For example where an individual submission has a material impact on the benchmark level. 31 Pg. 36, Par. 104 32 Article 3(1)(9) of the Regulation defines a submitter as a natural person employed by the contributor for the purpose of contributing input data. 12

ESMA stated that verifiability, whether it is applied to critical, significant or non-significant benchmarks [..], is not amenable to being implemented differentially. 33 We agree that even benchmarks lacking the systemic importance [..], may raise issues of transparency, market confidence, conflicts of interest, and investor protection. 34 However, we believe it will be crucially important for ESMA to allow for differential treatment of the principle of verifiability taking into consideration factors such as the number of contributors to the particular benchmark which would also be consistent with ESMA s stated objective to take into account the different types of benchmarks. 35 For example, imagine a benchmark with 100 contributors. On a given day, 95 of their submissions for a specific instrument were higher than on the previous day, while 5 contributors submissions were significantly below that of the previous day. We believe that it would be justified in such situation for the administrator to take a proportionate approach by focusing its efforts on verifying the input data from only those 5 contributors. We agree with ESMA s assessment that measures to ensure verifiability of input data should be imposed depending on the extent to which input data relies on expert judgement or the use of discretion. 36 Importantly, this would imply that different measures to ensure verifiability would be used for the various types of benchmarks depending upon the proportion of transaction data used in the benchmark. 5) Transparency of methodology (A7b) Q34: Do you consider the proposed list of key elements sufficiently granular to allow users to understand how a benchmark is provided and to assess its representativeness, its relevance to particular users and its appropriateness as a reference for financial instruments and contracts? We believe it is essential for administrators to provide users of their benchmarks with sufficient transparency about the benchmark objective as it will allow them to understand how a benchmark is provided and to assess its representativeness, its relevance to particular users and its appropriateness as a reference for financial instruments and contracts. That said, ESMA must carefully design any requirement to provide transparency of the benchmark methodology to not compromise the intellectual property of the benchmark owner or create a risk of front running. ESMA rightly recalls that these risks were already acknowledged in the context of the ESMA-EBA principles for benchmark-setting processes in the EU. 37 Specifically, they stated that transparency may be limited based on legal provisions safeguarding confidentiality and intellectual property rights. In addition, when a benchmark s methodology is fully disclosed, and input data is publicly available, there could be a risk of front-running. 38 ESMA should note that any failure to sufficiently protect intellectual property of benchmark owners in this context would result in a significant risk of stifling innovation in the sector. Q36: Do you agree that the proposed key elements must be disclosed to the public (linked to Article 3, para 1, subpara 1, point (a))? If not, please specify why not. We believe that users of a benchmark will generally need to be provided with a sufficient degree of transparency about the benchmark methodology. However, ESMA should consider that the information to be provided in this context is often confidential or protected and that benchmark owners are very concerned about the potential unintended consequences of disclosing certain aspects of their methodologies both for 33 Pg. 35, Par. 103 34 Pg. 35, Par. 103 35 Pg. 35, Par. 102 36 Pg. 36, Par. 104 37 https://www.eba.europa.eu/documents/10180/16145/2013-658+esma- EBA+Principles+on+Benchmarks+Final+Report.pdf 38 Pg. 40, Par. 118 13

competitive reasons and for potentially exposing their intellectual property. On the other hand, providing this information to the public rather than only to users creates little benefit. We therefore recommend that the required transparency about key elements of the methodology is provided to users of the benchmark. Q37: Do you agree with ESMA s proposal about the information to be made public concerning the internal review of the methodology? Please suggest any other information you consider useful to disclose on the topic. Consistent with our response to Question 36 above, we believe that, as long as the information concerning the internal review of the methodology is to be made available to users of the benchmark upon request, the disclosure requirement would be appropriate. Q39: Do you agree, in particular, on the opportunity that also the replies received in response to the consultation are made available to the public, where allowed by respondents? We agree with this as long as the publication of the replies received in response to the consultation does not breach any confidentiality obligations. 6) Code of Conduct (A9) Q43: Do you agree that a benchmark administrator could have a standard code for all types of benchmarks? If not, should there be separate codes depending on whether a benchmark is critical, significant or non-significant? Please take into account your answer to this question when responding to all subsequent questions. The Submitter Code of Conduct ( SCoC ) contains provisions on information provided by the contributors to the benchmark administrator with respect to input data, record keeping, suspicious input data reporting and conflict management requirements. 39 Given the large variety of benchmarks that likely to be subject to the Regulation, we urge ESMA to provide administrators with sufficient flexibility to decide on whether to establish a single or multiple SCoCs as long as they contain all of the provisions necessary to address the relevant requirements across all asset classes and benchmark type. To make the creation, publication and adherence to the SCoC an effective and efficient process, administrators should be permitted to choose to either establish provisions covering all of their benchmarks in a single SCoC or to create a number of separate SCoCs, distinguishing, for example, by asset class or benchmark type. We also propose that ESMA amend the requirement for administrators to continuously assess contributors adherence to the SCoC as administrators would find it operationally impossible to comply with such requirement. Instead it would be more practical for administrators to assess submitters compliance with the SCoC in regular intervals based on contributor representations of their compliance with the code. This could, for example, be performed on an annual basis. Furthermore, the SCoC would contain provisions on information to be provided by contributors to administrators. The structure of the SCoC should hence be at the discretion of the administrator depending on the nature of the benchmark. Q44: Do you believe that an administrator should be mandated to tailor a code of conduct, depending on the market or economic reality it seeks to measure and/or the methodology applied for the determi- 39 Pg. 46, Par. 138 14